| <<<Back 1 day (to 2013/11/20) | 2013/11/21 |
Robin_Watts | marcosw1: I am here. | 00:04.26 |
| Thanks. I will look into that tomorrow. If you have a list of them that would be great. | 00:04.48 |
marcosw1 | Robin_Watts: it's 2:00am, go to bed. | 00:04.50 |
Robin_Watts | It's just midnight here. | 00:05.01 |
marcosw1 | no wait, it's only midnight. | 00:05.01 |
| I'll open a bug with a list. | 00:05.20 |
Robin_Watts | marcosw1: Thanks. | 00:43.36 |
paulgardiner | Micha`: re test against SMALL_FLOAT in pdf_appearance.c: yes, well spotted. | 11:28.13 |
Robin_Watts | Crumbs, when marcos says empty file he really means an empty file. | 12:25.10 |
| not a blank page, but a file with 0 bytes in it. | 12:25.17 |
ray_laptop | kens: I saw your reply to cust 801. I'm wondering if we can use an interpreter "context switch". But certainly starting a separate process of gs would work. Suspending the multi-page job can be done in the gp_check_interrupts. If the 'pause' flag is set, just loop sleeping and checking until it is no longer set | 15:54.19 |
Robin_Watts | ray_laptop: I think using a second instance of gs is easier. | 15:54.44 |
kens | ray_laptop : Like Miles I hadn't intended to CC the customer :-( | 15:54.47 |
| Fortunately, like Miles, I hadn't written anything contentious, I was really 'thinking out loud' | 15:55.20 |
Robin_Watts | With a GS_THREADSAFE build, we can have multiple instances from the same DLL. | 15:55.26 |
ray_laptop | Robin_Watts: yes, better than trying to use an interpreter context | 15:55.27 |
kens | It seems the easier solutin to me, but it very much depends on their total system I htink | 15:55.52 |
ray_laptop | Robin_Watts: well, have we tested THREADSAFE enougn ? | 15:55.59 |
kens | ray_laptop : I suppose technically we could use the Display PostScript to do a context switch, that hadn't occured to me | 15:56.31 |
ray_laptop | They'll be running linux or windows so they could just have a process instead of relying on THREADSAFE | 15:56.47 |
kens | I think a separate process would be easier | 15:57.02 |
Robin_Watts | I think trying to use a context switch has the potential to cause us massive pain. | 15:57.10 |
ray_laptop | kens: and I'm not at all confident that context switching is well enough tested | 15:57.21 |
Robin_Watts | It *should* work, but when it doesn't, do you really want to have to try to fix it? | 15:57.22 |
kens | Robin_Watts : technically its all there I htink for DPS. | 15:57.25 |
| But I'd still prefer two processes :-) | 15:57.46 |
Robin_Watts | 2 processes or 2 instances, either is fine. | 15:58.12 |
kens | I seem to remember someone (luser_droog ?) muttering about DPS..... | 15:58.16 |
Robin_Watts | 2 instances may be easier for them. | 15:58.20 |
kens | Yep, I was treating the two interchangeably | 15:58.34 |
| Like I said I didn't really intend to CC the customer, better if those of you dealing with them continue to do so | 15:58.59 |
ray_laptop | kens: OK. np. you didn't say anything offensive | 16:02.13 |
kens | Fortunately :-) | 16:02.21 |
ray_laptop | At least they explained that they want to do this with PS input, so the 'easy' solution won't work. | 16:03.02 |
kens | yes, it was very helpful to see that. | 16:03.20 |
ray_laptop | I have to take my son to school. I'll be back online in a bit | 16:06.03 |
Robin_Watts | oh, bugger. | 16:14.45 |
kens | ? | 16:18.16 |
Robin_Watts | It's ok, I think. | 16:19.09 |
| I thought I'd spotted a big thinko in the alignment/padding work. | 16:19.21 |
| but I think it might just be a silly slip rather than a huge conceptual chasm. | 16:19.38 |
marcosw1 | henrys: you henrysx6 box seems to be having trouble running cluster jobs. I tried logging into it but can't reach your macpro, presumably related to your recent move? | 16:44.48 |
henrys | marcosw1: I'll have a look I upgraded henrysx6 | 16:45.50 |
| marcosw1:what is the hostname you are using for macpro | 16:46.35 |
| ? | 16:46.36 |
marcosw1 | henrys: henry-artifex.homeip.net | 16:47.11 |
henrys | marcosw1: funny the router stopped forwarding to that address | 16:51.39 |
| marcosw1: I've not really changed anything. | 16:52.17 |
| I see it is now forwarding ssh to the wrong machine, but I have no idea how it changed | 16:55.04 |
Robin_Watts | marcosw1: I think I've fixed both the SEGV and the empty files for psdcmykog now. | 16:56.05 |
marcosw1 | henrys: the only time I've ever had routers act strange is after a firmware update. | 16:56.10 |
| Robin_Watts: thx. I'll run another test immediately and see if there are any remaining problems. | 16:56.35 |
ray_laptop | kens: BTW, I looked at the driver some more from Martin (396) and it cannot possibly work with ADDITIVE and the devn configuration. | 16:57.01 |
kens | ray_laptop : I suspected this would be the case | 16:57.13 |
| I thought right form the off that was unlikely to work | 16:57.23 |
ray_laptop | kens: I actually built it to test some changes and found that it always fails with a put_params problem. The reason is that the devn_get_params returns "PageSpotColors", but the devn_put_params doesn't read that param for additive devices. | 16:58.25 |
Robin_Watts | ray_laptop: Have you pulled recently from the private repo? | 16:58.40 |
kens | That's probably why they got an error then | 16:58.41 |
Robin_Watts | ray_laptop: Like, in the last 2 hours? | 16:59.06 |
| If not, don't. | 16:59.12 |
ray_laptop | kens: I changed to subtractive and got reasonable looking output from p2 (100% black for the separation) | 16:59.17 |
| Robin_Watts: no, I haven't | 16:59.24 |
Robin_Watts | fab. I can safely redo the merge I just did. | 16:59.34 |
ray_laptop | Robin_Watts: OK, I won't | 16:59.35 |
| kens: but now tiger is all white :-( | 17:00.06 |
kens | Not good | 17:00.17 |
henrys | marcosw1: I've changed it but I may have to reboot the router | 17:00.23 |
| marcosw1: it still isn't going through | 17:00.40 |
Robin_Watts | ray_laptop: OK. I've merged master into the private branch. a couple of changes there that we should probably forward on to 801. | 17:01.00 |
marcosw1 | henrys: okay. | 17:01.27 |
henrys | bbiam | 17:01.33 |
ray_laptop | kens: so, I'll continue looking at why. I think it's because the ICC color handling assumes that 1 component is DeviceGray and doesn't call the device's color mapping procs | 17:01.45 |
| I suspect that there are *many* places in the code that assume 1 component is DeviceGray :-( And mvrhel isn't here to ask | 17:02.47 |
| Robin_Watts: is the private repo being kept up to date w.r.t. the origin/master ? | 17:03.43 |
Robin_Watts | ray_laptop: I have just pushed the latest master to it. | 17:04.16 |
ray_laptop | Robin_Watts: OK. Thanks | 17:04.26 |
marcosw1 | bbiab | 17:05.16 |
henrys | maybe I have to reset the cable modem also marcosw1 hang on | 17:15.23 |
| marcosw: the ip address on my router say 10. something but everything else maps to a 76. something address | 17:27.17 |
Robin_Watts | henrys: All your internal network IPs will be 10.0.0.x probably ? | 17:27.57 |
henrys | no they are all 192's | 17:28.17 |
Robin_Watts | oh, ok. | 17:28.36 |
| henry-artifex.homeip.net is currently 76.25.94.13 | 17:28.59 |
| which we would expect to be the external IP of your cable modem. | 17:29.09 |
marcosw1 | henrys: the last time the ddns updated for you was on Nov 12. Is that when you moved? | 17:29.27 |
henrys | Robin_Watts: there about's yes | 17:29.47 |
| oh marcosw1 ^^^ | 17:30.11 |
marcosw1 | np, i figured that was an answer to my question. | 17:30.32 |
Robin_Watts | henrys: Can you check (via a web thing maybe?) whether 76.25.94.13 is your current address? | 17:30.59 |
| http://whatismyipaddress.com/ | 17:31.15 |
henrys | in the router it says the internet ip address is 10.0.0.17 and it should be a 192 thing I'd guess. | 17:31.16 |
| Robin_Watts: yes I diid that and I get 76.* | 17:31.27 |
| I'm going to reboot the router again and see if this "internet port" doesn't change the 10.* is screwy | 17:33.22 |
kens | OK goodnight all | 17:34.19 |
henrys | the problem is I have a new cable modem, that is the 10.0.0.17 address and I think it is acting as a router - whereas my old cable modem was just a modem | 17:38.05 |
| marcosw1: do you want me to try and fix the cluster problem first? | 17:39.44 |
marcosw1 | henrys: I need to run out for a couple of hours so if you have time go ahead. otoh, it's not urgent, the cluster is happy running with the 22 remaining nodes (hubbles is down as well, with what appears to be a hardware failure). | 17:42.21 |
henrys | something to do with svn failing | 17:42.42 |
| marcosw: for the logs I'm pretty sure the node should work now. My networking situation is a bit more complicated. | 17:51.12 |
Robin_Watts | henrys: Probably, you just need to set up your cable modem to forward the appropriate port to your other router. | 17:52.04 |
henrys | Robin_Watts: yes that should work but I was thinking about getting rid of one of these routers | 17:52.31 |
Robin_Watts | The 'inner' router, presumably. | 17:52.44 |
henrys | or just using the cable modem router | 17:52.53 |
| or I'd prefer it act like a bridge | 17:53.22 |
Robin_Watts | right, that's what I meant, by getting rid of the "inner" router. | 17:53.36 |
henrys | Robin_Watts: oh right | 17:53.54 |
marcosw1 | henrys: I just tried a test clusterpush and henrysx6 failed again. | 18:07.38 |
henrys | I only ran svn upgrade on the test directory, have to do tests_private too. Any other test dirs? | 18:13.19 |
| I have to finish shoveling snow bbiam | 18:14.16 |
| marcosw1: it is still doing the upgrade .. maybe we should just get rid of svn ... | 18:27.53 |
| marcosw1: you can try it again or tell me how I can test it locally to see if there are other problems. | 18:40.47 |
| Robin_Watts: apparently you have to call the cable company to put the router in bridge mode, and many folks are saying they couldn't get things to work at all after that, so they went back. | 18:54.06 |
| so I'm hesitant to fool with it. | 18:55.23 |
Robin_Watts | henrys: It may not be an altogether bad thing to leave it as it is. | 18:55.33 |
| It means if you ever change the cable modem the rest of your network config is unaffected. | 18:56.09 |
henrys | as long as it supports forwarding which marcosw1 needs | 18:56.50 |
Robin_Watts | marcosw1: Any luck? | 20:50.25 |
| I mean, any indication from the test if it's working now? | 20:50.45 |
ray_laptop | darn. I tried getting bmpcmp results with extras=--saved-pages-test and it failed because the extras=... got fed to bmpcmp. I tried it with the extras=... BEFORE the bmpcmp and the dashboard shows bmpcmp *first*. I don't know if that's what''ll actually get run, but if so, it's a waste of time :-( | 21:49.12 |
marcosw | henrys: henrysx6 appears to be working now. | 21:50.54 |
| Robin_Watts: the psdcmyog device no longer generates empty files for nor core dumps, I'm running a comparison to psdcmyk now. | 21:52.00 |
| ray_laptop: unfortunately the --saved-pages-test option is still being passed to the bmpcmp command. I'll try to figure out how Robin_Watts intended extras= to work with bmpcmp. | 21:54.25 |
| ray_laptop: for the logs, I found the issue with bmpcmp and extras= but it will take me a bit to fix and test it. | 22:02.53 |
ray_laptop | marcosw: the parameters should be fed to gs, not to bmpcmp. | 22:20.10 |
| marcosw1: where is the cluster code git ? | 22:27.26 |
marcosw1 | ray_laptop: on casper in file:///home/git/cluster.git (don't ask me how to fetch it remotely, I only ever git push on casper). | 22:29.23 |
| there appear to be two problems with bmpcmp and extras= out of character for me I'm actually going to try and fix it correctly rather than apply a hack | 22:30.07 |
ray_laptop | marcosw1: OK. If you weren't, I was going to hack on it :-) | 22:30.46 |
| and given my .pl skills, it might have been pretty UGLY | 22:31.11 |
marcosw1 | looking at the code I'm not sure anything other than ugly is an option. | 22:31.47 |
ray_laptop | bbiaw -- have to pick my twins up from school... | 22:34.30 |
marcosw1 | ray_laptop: okay, I think it's working, I'll fire off a bmpcmp job for you. | 22:34.48 |
| with extras=--saved-pages-test | 22:34.55 |
| Forward 1 day (to 2013/11/22)>>> | |